Abstract—Real-time relative pose (RP) estimation is a corner-stone for effective multi-agent collaboration. When conventionalglobal positioning infrastructure such as GPS is unavailable, theuse of Ultra-Wideband (UWB) technology on each agent providesa practical means to measure inter-agent range. Due to UWB{\textquoteright}sprecise range measurements and robust communication capa-bilities, external hardware installations are not needed. However,when only a single UWB device per agent is used, the relative posebetween the agents can be unobservable, resulting in a complexsolution space with multiple possible RPs. This paper proposes anovel method based on an Unscented Particle Filter (UPF) thatfuses single UWB ranges with visual-inertial odometry (VIO).The proposed decentralized method solves the multi-modal solu-tion in 3D (4-DoF) for the RP when it is unobservable. Moreover,a pseudo-state is introduced to correct the rotational drift ofthe agents. Through simulations and experiments involving tworobots, the proposed solution was shown to be competitive andless computationally expensive than state-of-the-art algorithms.Additionally, the proposed solution provides all possible relativeposes from the first measurement. The code and link to the videoare available https://github.com/y2d2/UPF RPE.